The study protocol adhered to the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of Wenzhou Medical University. All patients have been informed of the purpose of the experiment and signed informed consent. Patients with ERMs were recruited from January to December 2021 at the Affiliated Eye Hospital of Wenzhou Medical University in Hangzhou.

Each patient underwent a complete ophthalmic examination, including dilated fundus examination by two experienced specialists (Shen Lijun and Cheng Dan), best-corrected visual acuity (BCVA) measurement using Snellen charts, intraocular pressure measurement (Goldmann applanation tonometry), and spectral-domain OCT (SD-OCT, Heidelberg Engineering, Heidelberg, Germany).

The inclusion criteria were as follows: patients with clinically diagnosed unilateral ERM who underwent surgical removal, with a minimum follow-up of three months. The exclusion criteria were as follows: eyes with other intraocular diseases such as uveitis, age-related macular degeneration, diabetic retinopathy, severe cataract, high myopia (> − 6 diopters), glaucoma, vitreomacular traction syndrome; eyes with previous vitreoretinal surgery; and OCT images of poor quality. No obvious postoperative complications occurred after surgery, such as, Irvine-Gass syndrome，macular hole, retinal detachment, and endophthalmitis. Moreover, considering the impact of complex surgery on the choroid, we excluded intraoperative retinal photocoagulation for retinal tears, holes or rhegmatogenous retinal detachments.

Patients and controls underwent a horizontal single-line EDI-OCT scan encompassing the fovea (Heidelberg Engineering, Heidelberg, Germany). EDI-OCT images were obtained between 9:00 AM and 12:00 AM to restrict the effect of the circadian rhythm on choroidal parameters [13]. The choroidal thickness (CT) and choroidal vascularity were assessed using the scan images. The choroid was defined as the area between the retinal pigment epithelium (RPE)–Bruch’s membrane complex and the choroid–sclera interface in SD-OCT images. After semiautomatic choroidal segmentation using a custom method built in MATLAB R2017a (MathWorks, Natick, MA, USA), a trained examiner (Ruan Kaiming) manually modified segments of the RPE–Bruch’s membrane complex and choroid–sclera interface. After segmentation, each image was binarized in MATLAB R2017a using custom-created algorithms to demarcate the LA and stromal area using Niblack’s auto local threshold, which was proposed by Sonoda et al. [14]. The mean macular CT, TCA, LA, and stromal area were estimated after image processing, and the size was adjusted to account for changes in magnification between the eyes due to differing ALs. The ratio of LA to TCA was used to calculate CVI. The region of interest was defined as a 6-mm macular region centered on the fovea.

The macular zone was split into three concentric rings with diameters of 1 (central fovea, C), 3 (parafovea), and 6 mm (perifovea). The central region centered on the fovea had a diameter of 1 mm. N1 and N2 nasal regions extended from an inner diameter of 1 mm and 3 mm to an outer diameter of 3 mm and 6 mm, respectively. T1 and T2 temporal regions extended from an inner diameter of 1 mm and 3 mm to an outer diameter of 3 mm and 6 mm, respectively. CT and CVI in each region were calculated using horizontal B-scans (Fig. 1).

Choroidal vascularity index (CVI) and choroidal thickness were measured in Macular according to ETDRS grid (A) Showing steps (B1-4) using semi-automatic algorithms in MATLAB R2017a. T1, temporal parafovea; T2, temporal perifovea; C, center; N1, nasal parafovea; N2, nasal perifovea

In addition, according to Govetto et al., we categorized ERM into four stages using B-scan images across the macula [15]. Briefly, stage 1 is defined as the presence of the foveal pit and well-defined retinal layers; stage 2 is defined as the absence of the foveal pit and well-defined retinal layers; stage 3 is defined as the absence of the foveal pit and well-defined retinal layers, and the presence of ectopic inner foveal layers; stage 4 is defined as the absence of the foveal pit, presence of ectopic inner foveal layers, and disrupted retinal layers.

All procedures were performed by a single surgeon (Shen, Lijun). All 65 epiretinal membrane eyes received retrobulbar anesthesia preoperatively. We dotted a mark on the sclera at a distance of 3.5 mm from the corneoscleral limbus with an angle ruler. The trocar was first inserted through the sclera with a 30-degree angle and then oriented tangentially to the sclera. The conjunctiva and sclera were penetrated with the trocar, and a microcannula was implanted. Standard 3-port pars plana vitrectomy (PPV) was performed using 23-gauge instruments (Constellation Vision System, Alcon Laboratories, Fort Worth, Texas, USA). Cataract surgery was performed in all 65 phakic eyes. A 0.025% indocyanine green solution (DANDONG YICHUANG PHARMACEUTICAL CO., LTD) was injected into the vitreous cavity and washed out no later than 3 s using a 23-gauge cutter. The ERM and inner limiting membrane was peeled by careful grasping with forceps (Grieshaber Revolution DSP, ILM Forceps, Alcon) without damaging the retina to approximately 4-disc diameters centered on the macula. In 65 epiretinal membrane eyes, complete posterior vitreous detachment (PVD) accounted for the majority of epiretinal membrane eyes (n = 52), and the remaining few (n = 13) were incomplete PVD and induced intraoperatively. All patients underwent fluid–air exchange, and the sclerotomies were sutured by 8-0 polypropylene. And there was no sign of leakage. No complications due to leakage such as low IOP occurred postoperative. Patients were instructed to remain in the prone position for 3–5 days postoperatively. Within one month after epiretinal membrane surgery, we routinely used levofloxacin eye drops 4 times a day and tobramycin-dexamethasone 4 times a day at first week and decreasing once a week. None of these drugs have been reported to have an effect on CVI. No medication for controlling IOP were used.

Statistical analysis was performed using Statistical Package for the Social Sciences statistical software (SPSS Inc., Chicago, IL, United States). The mean with standard deviation for the normal distribution is shown in the tables. BCVA was converted to the logarithm of the minimum angle of resolution (logMAR) for data analysis. For logMAR of BCVA, paired t-tests were applied. To compare CVI and CT between eyes with ERM and fellow eyes, paired t-tests were used. To compare CVI and CT at 1 day before surgery with those at 7, 30, and 90 days after surgery, repeated-measures ANOVA tests were applied, and post hoc analyses were performed. Statistical significance was set at p < 0.05.